How AI describes Florence-2 Large

Florence-2-large is a Microsoft vision foundation model on Hugging Face that uses a prompt-based, sequence-to-sequence architecture to perform vision and vision-language tasks including captioning, object detection, segmentation, and OCR. Trained on the FLD-5B dataset (5.4B annotations across 126M images), it supports zero-shot and finetuned performance and features a 4k context length. The model family includes base and large variants and finetuned versions, with example code and tutorials for loading and inference via Hugging Face transformers.